Visual Abstract
Abstract
Psychedelic compounds such as 3,4-methylenedioxymethamphetamine (MDMA) have attracted increasing interest in recent years because of their therapeutic potential in psychiatric disorders. To understand the acute effects of psychedelic drugs in vivo, blood-oxygenation-level–dependent (BOLD) functional MRI (fMRI) has been widely used. In particular, fMRI studies have suggested that MDMA leads to inhibition of brain activity, challenging previous hypotheses indicating mainly excitatory effects based, among others, on increased metabolism shown by 18F-FDG functional PET (fPET). However, interpretation of hemodynamic changes induced by psychedelics is difficult because of their potent vascular effects. Methods: We aimed to delineate the acute effects of MDMA using simultaneous PET/fMRI in rats. For this purpose, hemodynamic changes measured by BOLD fMRI were related to alterations in glucose utilization and serotonin transporter (SERT) occupancy using 18F-FDG fPET/fMRI and 11C-DASB PET/fMRI. Results: We show that MDMA induces localized increases in glucose metabolism in limbic projection areas involved in emotional processing. The increased glucose metabolism was accompanied by global cerebral and extracerebral hemodynamic decreases. We further demonstrated a strong correlation between SERT occupancies and regional BOLD reductions after acute MDMA administration. Conclusion: Our data indicate that hemodynamic decreases after acute MDMA administration are of a nonneuronal nature and initiate peripherally. Within the brain, MDMA triggers neuronal activation in limbic projection areas, whereas increased serotonin levels induced by SERT blockage cause neurovascular uncoupling through direct vascular effects. Correct understanding of the in vivo mechanism of MDMA not only supports ongoing research but also warrants a reassessment of previous studies on neuronal effects of psychedelics relying on neurovascular coupling and recommends 18F-FDG fPET as a potentially more robust measure for pharmacologic research.
Psychedelic drugs, including lysergic acid diethylamide, psilocybin, and 3,4-methylenedioxymethamphetamine (MDMA), have recently gained increasing attention because of their potential benefits for treating psychiatric disorders (1). MDMA-assisted psychotherapy is currently in a phase 3 clinical trial to treat severe posttraumatic stress disorder, with encouraging initial results (2). Research in this area is also increasingly associated with the development of imaging techniques as quantitative biomarkers in addition to behavioral parameters (1). To investigate the mechanisms of psychedelic drugs in vivo, MRI methods inferring neuronal activity through neurovascular coupling, such as blood-oxygenation-level–dependent (BOLD) functional MRI (fMRI) and arterial spin labeling, have been widely used (3–5). Interestingly, research performed over the last decade using these methods has shown that psychedelic compounds such as MDMA (3) and psilocybin (4) inhibit brain activity, contradicting previous studies that indicated mainly excitatory effects (6–9).
However, the use of hemodynamic methods may be insufficient to understand the effects of psychedelics on neuronal activity. First, psychedelic drugs elicit their effects by strongly affecting one or more neurotransmitter systems (10). Thus, it is crucial to evaluate hemodynamic changes in relation to neurotransmitter alterations. Second, in addition to neuronal effects, an increase in neurotransmitters such as serotonin elicited by psychedelic compounds can have potent vascular effects (11–13). This aspect is particularly critical for methods based on neurovascular coupling, such as BOLD fMRI and arterial spin labeling. The emergence of hybrid PET/MRI allows simultaneous assessment of brain function at multiple physiologic levels. The combination of PET with pharmacologic MRI can offer important complementary insight on drug mechanisms (14,15). Furthermore, recent developments in the administration of 18F-FDG PET via constant infusion (16) have paved the way toward functional PET (fPET). fPET enables the imaging of changes in glucose metabolism at a resolution of minutes (17), providing a more robust indirect measure of neuronal activity than is possible with fMRI, being largely immune to vascular changes (16).
We aimed to exploit the potential of multimodal imaging to characterize the acute effects of MDMA using PET/fMRI. First, we performed 18F-FDG fPET/fMRI to simultaneously determine hemodynamic and metabolic changes elicited by MDMA and thereby elucidate potential inhibitory or excitatory actions of this compound. In a second cohort, we used 11C-3-amino-4-(2-dimethylaminomethylphenylsulfanyl)-benzonitrile (11C-DASB) to investigate relationships between hemodynamic alterations and changes in serotonin transporter (SERT) availability, one of the main targets of MDMA (18).
MATERIALS AND METHODS
Animals
Male Lewis rats (n = 29) were obtained from Charles River Laboratories and divided into 2 groups: 18F-FDG fPET/fMRI was performed on 17 animals (body weight, 361 ± 19 g), whereas 11C-DASB PET/fMRI was performed on 11 animals (body weight, 365 ± 19 g). Nine fMRI datasets were excluded from the study because of motion during acquisition. One 11C-DASB PET and two 18F-FDG fPET datasets were excluded from the analysis because of paravenous tracer injections. The animals were kept at a room temperature of 22°C and 40%–60% humidity under a 12-h light–dark cycle. The rats were fed with standard diet and received tap water ad libitum. They were kept fasting for 6 h before the start of the experiments. All experiments were performed in accordance with the German Federal Regulations on the Use and Care of Laboratory Animals and approved by the Tuebingen regional council. Two additional cohorts scanned under the same 18F-FDG fPET/fMRI and 11C-DASB PET/fMRI protocols, but exposed to phosphate-buffered saline instead of MDMA, are presented in the supplemental materials (available at http://jnm.snmjournals.org) (3–6,10,11,15,16,19–36).
Simultaneous PET/fMRI Experiments
The animals were scanned under 1.3% isoflurane and constant monitoring of breathing rate and temperature (Supplemental Fig. 1) using a 7-T small-animal MRI scanner (ClinScan; Bruker). T2-weighted anatomic reference scans and fMRI scans (repetition time, 2,000 ms; echo time, 18 ms) were obtained using a linearly polarized radiofrequency coil for transmission and a 4-channel surface rat brain coil for reception. The PET scans were acquired simultaneously using an in-house–developed insert and reconstructed into 100 frames of 1 min using an ordered-subsets expectation-maximization 2-dimensional algorithm. The MDMA challenge (3.2 mg/kg) was applied 40 min after the start of the acquisition. Additional details on the experimental procedure are provided in the supplemental materials.
Data Analysis
Statistical Parametric Mapping (SPM12, Wellcome Trust Centre for Neuroimaging) via Matlab (The MathWorks) and Analysis of Functional NeuroImages (AFNI, National Institute of Mental Health) were used for data preprocessing as previously reported (37). An extensive description of all preprocessing and analysis steps can be found in the supplemental materials. Average time courses were extracted from all datasets after preprocessing using the MarsBaR toolbox (38) and regions of interest (see Fig. 1 for abbreviations and Supplemental Table 1 for their respective volumes) defined by the atlas of Schiffer et al. (34). Additionally, extracerebral BOLD fMRI signals were extracted using binary masks generated with AFNI. The general linear model available in SPM was applied to determine voxels with significantly altered fMRI and PET signals after MDMA exposure. For all datasets, baseline was defined as 30–40 min after the scan start, when tracer equilibrium had been reached between the regions with high SERT density and high 18F-FDG uptake and the reference regions. The 18F-FDG fPET data were normalized using cerebellar uptake. For 11C-DASB PET, the cerebellar gray matter was chosen as the reference region as previously described (36). For all methods, the signal at baseline was compared with six 10-min blocks between the challenge (40 min after the scan start) and the end of the scan (100 min after the scan start). Group-level T-maps were generated for all cohorts, methods, and time periods. All T-maps were subjected to voxelwise signal quantification to determine the regional contributions of the brain regions selected. The average T-scores of all voxels comprising each region were calculated for each period and modality to compare the respective spatial patterns of MDMA effects on hemodynamics, glucose metabolism, and SERT occupancy.
Regionwise evaluation of 18F-FDG fPET and BOLD fMRI signal changes. (A) Time–activity curves for all regions and whole-brain average. Voxelwise normalized uptake maps indicate 18F-FDG uptake at baseline and at 90–100 min. (B) Regional BOLD fMRI signals normalized to respective baseline periods. (C) Both signals normalized to baseline period over last 10 min before MDMA administration for common frame of reference. Amyg = amygdala; CPu = caudate putamen; Ins = insular cortex; MB = midbrain; mPFC = medial prefrontal cortex; WB = whole brain.
RESULTS
Metabolic Increases Accompany Hemodynamic Reductions
First, we investigated the relationship between hemodynamic and metabolic changes after acute MDMA administration using simultaneous 18F-FDG fPET/fMRI (Figs. 1 and 2; Supplemental Fig. 2).
General-linear-model analysis of 18F-FDG fPET/fMRI cohort. (A) Voxelwise analysis of both signals at 90–100 min (P < 0.05 family-wise error–corrected at voxel level for PET, P < 0.001 at voxel level with P < 0.05 family-wise error cluster-level correction for fMRI, n = 15 for fPET, n = 9 for fMRI). (B) Bar diagram indicating average T-scores for each region and modality. (C) Average regional T-scores plotted in scatter diagram to evaluate spatial correlation of both readouts. Amyg = amygdala; Au = auditory cortex; CAad = anterodorsal hippocampus; CAp = posterior hippocampus; Cg = cingulate cortex; CPu = caudate putamen; Ent = entorhinal cortex; GLM = general linear model; Hypo = hypothalamus; IC = inferior colliculus; Ins = insular cortex; M1 = motor cortex; MB = midbrain; mPFC = medial prefrontal cortex; NAc = nucleus accumbens; OC = olfactory cortex; OFC = orbitofrontal cortex; PAG = periaqueductal gray matter; Par = parietal cortex; RS =retrosplenial cortex; S1 = somatosensory cortex; SC = superior colliculus; Sep = septum; Th =thalamus; V1 = visual cortex; VTA = ventral tegmental area.
The normalized 18F-FDG fPET time–activity curves for all regions and the average whole-brain time–activity curve and voxelwise uptake maps (Fig. 1A) indicated an increase in metabolism in the midbrain and in the anterior subcortical and frontal cortical areas, whereas more minor or no changes occurred in the posterior cortical regions. Notably, we found a simultaneous decrease in hemodynamics, as indicated by BOLD fMRI (Fig. 1B). The whole-brain–averaged BOLD fMRI signal was reduced by 4.5% 15 min after the challenge. Importantly, the data revealed that the decreases were global and occurred in all regions investigated. A temporal comparison of the 18F-FDG fPET and BOLD signal changes relative to baseline is shown in Figure 1C. The highest metabolic increases occurred in frontal areas, including the caudate putamen (22%), insular cortex (21%), medial prefrontal cortex (18%), and amygdala (13.5%). Increases in all regions (>1%) were observed within 5 min of the challenge.
The voxelwise general-linear-model analyses presented in Figure 2 revealed metabolic increases across several subcortical areas and in frontal cortical areas between 90 and 100 min. The medial prefrontal cortex and orbitofrontal cortex (T = 7.6 for both), along with the midbrain (T = 7.4), thalamus (T = 7.9), and nucleus accumbens (T = 7.4), exhibited the most significant 18F-FDG increases. The most significant BOLD fMRI decreases occurred in posterior areas such as the midbrain (T = 4.8), ventral tegmental area (T = 5.0), hypothalamus (T = 5.8), and pons (T = 5.9). The T-scores of metabolic increases and hemodynamic decreases did not correlate significantly (r = 0.21).
SERT Occupancy Changes Induced by MDMA Correlate with BOLD Decreases
To further elucidate the molecular underpinnings of the observed hemodynamic decreases, we evaluated BOLD fMRI changes concurrently with alterations in SERT availability using 11C-DASB PET/fMRI in a second cohort (Figs. 3 and 4; Supplemental Fig. 3).
Regionwise evaluation of 11C-DASB PET and BOLD fMRI signal changes. (A) Dynamic BPND for all regions and whole-brain average. Voxelwise maps indicate 11C-DASB binding at baseline, as well as 70–80 min after scan start. (B) Regional BOLD fMRI signals normalized to respective baselines. (C) Temporal comparison of PET and BOLD signal changes (normalized to baseline) in caudate putamen, medial prefrontal cortex, and midbrain. CPu = caudate putamen; DVR = distribution volume ratio; MB = midbrain; mPFC = medial prefrontal cortex; WB = whole brain.
General-linear-model analysis of 11C-DASB PET/fMRI cohort. (A) Voxelwise analysis of both signals at 70–80 min. Voxelwise maps are presented (P < 0.05, voxel-level family-wise error correction for PET and P < 0.001 at voxel level with P < 0.05 family-wise error cluster-level correction for fMRI, n = 11). (B) Bar diagram indicating average T-scores for each region and modality. (C) Regional T-scores plotted in scatter diagram to evaluate spatial correlation of both readouts. ***P < 0.001. Amyg = amygdala; Au = auditory cortex; CAad = anterodorsal hippocampus; CAp = posterior hippocampus; Cg = cingulate cortex; CPu = caudate putamen; Ent = entorhinal cortex; GLM = general linear model; Hypo = hypothalamus; IC = inferior colliculus; Ins = insular cortex; M1 = motor cortex; MB = midbrain; mPFC = medial prefrontal cortex; NAc = nucleus accumbens; OC = olfactory cortex; OFC = orbitofrontal cortex; PAG = periaqueductal gray matter; Par = parietal cortex; RS = retrosplenial cortex; S1 = somatosensory cortex; SC = superior colliculus; Sep =septum; Th = thalamus; V1 = visual cortex; VTA = ventral tegmental area.
The 11C-DASB nondisplaceable binding potential (BPND) reached equilibrium 30 min after injection (Fig. 3A). After the challenge, binding in all regions decreased either immediately (1–2 min after the challenge) in areas with high binding values (BPND > 1.8) or with a delay of approximately 10 min in regions with lower 11C-DASB binding values. At 30 min after the challenge, binding remained stable until the end of the scan. Similarly to the 18F-FDG fPET/fMRI cohort, all regional BOLD signals decreased within 6 min after the challenge (Fig. 3B). Regions with higher baseline SERT availability showed a faster response than regions with lower baseline SERT availability (Fig. 3C). For example, 11C-DASB binding in the midbrain (BPND = 2.1) decreased immediately after the challenge. In contrast, 11C-DASB binding in the caudate putamen (BPND = 1.6) and medial prefrontal cortex (BPND = 1.6) remained stable or increased briefly. After approximately 10 min, 11C-DASB binding decreased in all regions until it reached equilibrium at 30–40 min after the challenge (39% decrease for medial prefrontal cortex, 44% decrease for caudate putamen). BOLD decreases occurred homogeneously, peaking 30 min after the challenge (∼8.5% in medial prefrontal cortex, ∼5.5% in caudate putamen, and ∼6.5% in midbrain).
Figure 4 shows the voxelwise decreases at 70–80 min after the scan start. Regional averages showed the strongest decreases in the ventral tegmental area (T = 13.7), periaqueductal gray matter (T = 12.8), and midbrain (T = 12.7) for 11C-DASB and in the hypothalamus (T = 7.8), ventral tegmental area (T = 7.2), and thalamus (T = 6.9) for BOLD fMRI. Remarkably, regional T-scores of both readouts correlated strongly (r = 0.79, P < 0.001).
Hemodynamic Reductions Also Occur in Nonneuronal Tissues
The limited spatial extent of hemodynamic changes compared with the metabolic and SERT occupancy alterations may be due to the smaller magnitudes of the BOLD decreases. To further clarify this aspect, we merged the fMRI scans from both cohorts (Fig. 5). We also extracted the BOLD signals from extracerebral areas to investigate whether the BOLD decreases are specific to neuronal tissue.
General-linear-model analysis of BOLD decreases after merging BOLD fMRI datasets acquired in both cohorts (n = 20). Data are shown at P < 0.05, family-wise error–corrected at voxel level. Arrows indicate extracerebral decreases. Average BOLD signals were extracted from extracerebellar regions and plotted along with whole-brain signal. WB = whole brain.
After the cohorts were merged, the BOLD fMRI decreases were widespread and, intriguingly, also occurred in nonneuronal areas (Fig. 5). The average extracerebral BOLD signal decreased coherently with the cerebral BOLD signal, both reaching maximum reductions 27 min after the challenge (cerebral BOLD, 4.8%; extracerebral BOLD, 7.4%).
Additional Analyses
Analyses of the phosphate-buffered saline cohort and a comparison with the readouts shown after MDMA application are provided in Supplemental Figures 4–5. Furthermore, we extracted general-linear-model alterations induced by MDMA in each subject of both the 11C-DASB cohort and the 18F-FDG cohort to demonstrate the feasibility of subject-level PET inferences using our approach (Supplemental Figs. 6–7). Additionally, we reproduced the metabolic changes using whole-brain normalization and validated the choice of the cerebellum for normalization (Supplemental Fig. 8). Moreover, we examined the robustness of our correlation between hemodynamic and SERT decreases by using β-values and subject-level general-linear-model readouts (Supplemental Fig. 9). Finally, we compared the temporal characteristics of the observed hemodynamic, metabolic, and SERT availability changes (Supplemental Fig. 10).
DISCUSSION
Our data indicate that increased neuronal activity after MDMA is accompanied by neurovascular uncoupling, possibly mediated through the vascular effects of serotonin after SERT blockage.
Simultaneous Uncoupling Between Metabolism and Hemodynamics
Studies measuring glucose utilization after psychedelic challenges have shown mixed results, yet the main result has been increased metabolism (6,7,9,35), suggesting neuronal activation. However, more recent work performed by Carhart-Harris et al. (3), using arterial spin labeling challenged this hypothesis and argued that brain activity is exclusively decreased solely under MDMA. Our study confirmed that this finding also holds true in rodents. Carhart-Harris et al. speculated that MDMA exerts inhibitory effects directly through serotonin receptor (5-HT) 1A (3,39). Comparable reductions under acute psilocybin reported by the group of Carhart-Harris were attributed to the inhibitory effects of 5-HT2A receptors (4). On a side note, the hemodynamic decreases for MDMA and psilocybin in humans were located predominantly in the right hemisphere (3,4), similarly to our study, thereby supporting the translatability of our readout. Notably, Carhart-Harris et al. argued that possible discrepancies with previous work indicating increased metabolism using 18F-FDG PET (9) might be due to its inferior temporal resolution (4). Therefore, the authors claimed that the reported increases may represent a rebound in glucose metabolism after the acute inhibition captured by fMRI (4). We agree that earlier 18F-FDG PET or ex vivo studies (6,7,9,35) measuring cerebral glucose utilization lacked temporal specificity because of methodologic limitations on their interpretability in terms of acute effects. An additional confound is that the studies compared different, relatively small cohorts having received either placebo or MDMA. The present work overcomes all these limitations. The constant-infusion protocols used for both tracers enabled delineation of pharmacologic effects immediately after the challenge at 1-min intervals, simultaneously with hemodynamic alterations, and within the same subjects. Therefore, we demonstrated that the uncoupling between flow and metabolism previously shown (6,11,35) does occur simultaneously within the same subjects.
Origin of Peripheral and Cerebral Hemodynamic Decreases
We showed that nonneuronal effects dominate hemodynamic changes induced by MDMA. In particular, our data shed light on 2 separate phenomena. First, the temporal coherence between hemodynamic reductions in cerebral and extracerebral areas suggests that vascular effects occur at the periphery. The 5-HT2A receptor, which is postulated—along with the 5-HT1B receptor—to mediate vasoconstrictive effects (40,41), is one of the main targets of MDMA (5,42). However, because MDMA has a much stronger affinity to the 5-HT2A receptor than to the 5-HT1B receptor (5), direct agonist action of MDMA at 5-HT2A in peripheral blood vessels likely drives our results (40). Previous work has demonstrated the role of 5-HT2A in vasoconstriction of the carotid artery, the main vessel supplying blood to the brain (43,44). In addition to peripheral effects, the cerebral hemodynamic decreases are also likely of a serotonergic nature. Existing publications have indicated that serotonin impacts brain hemodynamics (12,13) and that direct manipulation of the raphé nuclei constricts cerebral microvasculature (11). In contrast to the peripheral hemodynamic reductions, which can be attributed solely to the direct effects of MDMA, the decreases in the brain may additionally be triggered by increased synaptic serotonin levels after SERT blockage. This finding is supported by the high correlation between SERT blockage and hemodynamic decreases in BOLD fMRI, suggesting that increased levels of endogenous serotonin may modulate cerebral hemodynamics. The exact involvement of different receptors needs to be investigated, for instance by combining psychedelic challenges with respective antagonists. Our results warrant a reevaluation of previous data (3,4) and generally call for caution when interpreting findings relying on neurovascular coupling under pharmacologic challenges (3,5,42).
Anatomy and Physiology of Increased Metabolism
We demonstrated that MDMA increases the metabolism of different regions, likely because of neuronal activation, as fPET has been shown to reliably map onto neuronal activity while being independent of hemodynamic changes (16). Interestingly, the metabolic increases were more weighted toward serotonergic projection areas than toward the midbrain regions showing the strongest reductions in SERT availability. First, this finding is in line with the hypothesis that most of the glucose is consumed postsynaptically (45). Second, the areas showing increased metabolism are consistent at a functional level with most previously reported behavioral effects of MDMA. The signals observed in the nucleus accumbens, amygdala, and insula align well with salience changes known from imaging and behavioral studies (3,46). In particular, the nucleus accumbens is involved in responses to numerous drugs (47). The amygdala, insula, and orbitofrontal cortex are strongly involved in emotional processes (48). Activity in the olfactory cortex and olfactory bulb could indicate increased food-seeking or sexual arousal (49,50). Enhanced metabolic activity in the sensory cortices is in concordance with heightened sensations elicited by MDMA (49). Furthermore, the 5-HT2A receptor exhibits a strong anterior–posterior gradient in the cortex, being strongly expressed in frontal areas of the cortex (51) and overlapping with the activations indicated by fPET, and has been shown to be responsible for serotonergic activation in projection areas such as the prefrontal cortex (52).
Other factors may play a role in the findings, and there are also certain limitations that need to be considered when contemplating our data. The supplemental materials provide a thorough discussion of these aspects.
CONCLUSION
The present study showed the potential of multimodal imaging in drug research. We demonstrated the acute neurovascular uncoupling induced by MDMA, characterized by increased neuronal activity in monoaminergic projection areas and accompanied by vascular depression of a serotonergic nature. Our results provide important insight into the mechanism of action of MDMA and pave the way for the application of 18F-FDG fPET and hybrid fPET/fMRI in drug research.
DISCLOSURE
This research was supported by funds from the Eberhard Karls University Tübingen Faculty of Medicine (fortüne 2209-0-0 and 2409-0-0) to Hans Wehrl, from the Carl Zeiss Foundation to Kristina Herfert, and from the Werner Siemens Foundation to Bernd J. Pichler, as well as by an international exchange grant from the Osaka Medical Research Foundation for Intractable Diseases to Tadashi Watabe. No other potential conflict of interest relevant to this article was reported.
KEY POINTS
QUESTION: What are the effects of acute administration of MDMA on neuronal activation in the brain?
PERTINENT FINDINGS: Global decreases in BOLD fMRI are of a vascular, rather than a neuronal, nature and strongly correlate with SERT occupancy measured simultaneously using 11C-DASB. In contrast, 18F-FDG fPET indicates simultaneous increases in limbic glucose consumption, potentially mapping onto neuronal activation.
IMPLICATIONS FOR PATIENT CARE: The study emphasizes the caveats of BOLD fMRI and other hemodynamic methods when strong vascular effects are present and recommends the use of 18F-FDG fPET as an alternative for tracking neuronal activity in vivo after pharmacologic challenges in drug research.
ACKNOWLEDGMENTS
We acknowledge Julia Mannheim, Rebecca Rock, Neele Hübner, Andreas Dieterich, Ines Herbon, Stacy Huang, Sandro Aidone, Linda Schramm, and the Radiopharmacy Department for their administrative and technical support. The graphical abstract was generated using BioRender. This work is part of the PhD thesis of Tudor M. Ionescu.
Footnotes
Published online Sep. 29, 2022.
- © 2023 by the Society of Nuclear Medicine and Molecular Imaging.
REFERENCES
- Received for publication May 26, 2022.
- Revision received September 14, 2022.
